United States
Environmental Protection
 Environmental Sciences Research
 Research Triangle Park NC 27714^, j
 Research and Development
 EPA-600/S4-81-034 Aug. 1981
Project  Summary
Case  Studies  in  the
Application  of  Air  Quality
Modeling  in  Environmental
Decision   Making
Catherine G. Miller
  Eleven case studies of the applica-
tion of air quality models were under-
taken in order to examine the problems
encountered when trying to use these
models in making environmental policy
decisions. The case studies of air
pollution control decisions describe
the decision process, the models used,
the critiques of the models, and the
participation by outside interest groups
in the decision process. The studies
include two cases of federal decisions,
seven state decisions, one local deci-
sion, and a review of the evolution of
modeling requirements in the Clean
Air legislation. The time covered is
from 1970 to the present and includes
several cases for which a final decision
has not yet been made.
  The results of this investigation
show that indeed  the well-known
technical and political constraints
exist but that unresolved policy issues,
the management  of the decision
process and conflicting institutional
and organizational interests also cause
problems. Recommendations are made
on how to improve the technical plan-
ning and management of the decision
process so that the air quality models
can become a better policy tool within
the state-of-the-art, political and or-
ganizational constraints.
  This Project Summary was devel-
oped by EPA's Environmental Sciences
Research Laboratory, Research Tri-
angle Park, NC, to announce key find-
ings of the research project that is fully
documented in a separate report of the
same title (see Project Report ordering
information at back).

  The Clean Air Act provides for  the
establishment of the ambient air quality
standards for  ubiquitous air pollutants
arising from  multiple sources. The
strategy of the legislation for achieving
these standards is emissions and land-
use controls. The jurisdiction for estab-
lishing these controls is shared between
the U.S. Environmental Protection
Agency (EPA) and various state and
local government agencies. This legisla-
tive framework, then, provides for  the
application of air quality models, mathe-
matical models which relate the emis-
sions of pollution sources to the pollutant
concentrations in the ambient air, as an
aid  in establishing the emissions and
land-use related controls.
  This research project consists of
eleven case studies of specific applica-
tions of air quality models by EPA and
state and local agencies. It grew out of
the concern that, although the legislation
and EPA regulations seem to call for the
use of air quality models in making
environmental control decisions, many
modelers as well as potential users of
models are dissatisfied: the modelers
because they are not listened to and the

users because they do not hear much
that they want to  listen to. Also, the
users may apply the models  in ways
other than those intended by the mod-
elers. Thus, the case studies examine
this conflict  between modelers and
decision makers (users). The focus is on
the decision process and what role the
technical information produced by the
models plays within that process.

  The studies reveal the problems that
arise in using research tools as policy
tools. In making environmental policy
decisions EPA and the state and local
agencies require  a sound scientific
basis. Scientific research, including air
quality models, has become increasingly
important as well as more sophisticated
and widespread. However, as  a useful
input to the decision making  process,
the research has in many cases not met
expectations. One problem often pointed
out is the state-of-the-art of environ-
mental research; such research has not
yet reached the level of maturity where
it can supply  information in the detail
and  accuracy required  by policy prob-
lems. Gaps certainly do exist in the
available information. However, the
reasons why these gaps exist and why
the flow of information between re-
searchers and decision makers is so
difficult can also be traced to the dif-
ferent professional and organizational
settings within which researchers and
decision makers work. Thus, the case
studies describe the decision  process,
what role the models played within this
political context and how the modelers,
given their professional and organiza-
tional orientation, did or did not respond
to the policy dimensions of the problem.
Except to outline the implications of the
parameters and assumptions chosen
for each model analysis, and the contro-
versies surrounding these  choices,  no
technical critique of the models them-
selves is made. Instead the case studies
focus on the uses of the models within
the decision process.

  The details of the case studies show
that the controversies and delays arise
not only from  the  model formulations
and capabilities but also from the process
as laid down  in the legislation and as
managed by EPA and the states. These
factors together cause a narrow and
sometimes contradictory definition of
both the problem to be resolved and the
solution proposed. No one factor stands
out as the dominant cause or easy target
for  reform.  Solutions appear to  be as
complex as the problem.
The Case Studies
  The case studies were picked from the
point of view of illustrating different
levels of decision making (national to
local), different institutional domains
(legislative, administrative and judicial)
as well as the different EPA programs
which require air quality modeling

Clean Air Legislation
  The overall strategy of the Clean Air
Act is one of technology-forcing, requir-
ing technical fixes which the experts do
not necessarily agree are available. This
strategy was inaugurated in 1970 when
the Congress concluded that  if they
waited until the technology was available
and control was demonstrated to be
economically feasible, it  would never
happen. Thus, they constructed a system
of national ambient standards based on
health effects but not cost considera-
tions. Such a system depends,  in part,
on having air quality  models to predict
the relationship between  source emis-
sions and ambient quality.
  In 1977 the Clean Air Act was amended
to include modeling requirements in
three of  its  provisions: prevention of
significant deterioration (PSD), visibility
protection, and nonattainment. During
the legislative debate on these amend-
ments the  accuracy and predictive
capabilities of the models was discussed
but generally ignored, because no alter-
natives were proposed.  The general
agreement was that models should be
only one policy-making tool but little
attention was given to what other tools
might be available. Instead attention
was focused on the political questions
of how to strike  the  balance between
protection of health and  economic
feasibility. To the extent that EPA might
be tempted too far to the economic side,
the modeling requirement was inserted
to focus efforts on the standards.
  The technology-forcing philosophy
and standard-setting strategy of the act
reinforced the tendency not to investigate
too deeply whether models to suit the
new programs were available. If the
health-based standards require a given
level  of control, then technology to
achieve that level will have to be devel-
oped. Similarly if the strategy of stan-
dards implies the use of modeling, then
models will be required.
  Thus, the legislators naturally debated
the political questions, leaving the
implementation question to the admin-
istrators. They have legitimized the use
of the air quality models while not
actually having applied the models or"
made decisions requiring the use of the
models. As the other case studies show,
the Clean Air  Act modeling provisions
have been interpreted as requiring
modeling in almost every case and have
occasioned much controversy and delay
generated by special interests (both
industry and environmentalists), situa-
tions possibly not envisioned by the

  In 1977 Harvard University applied to
the state of Massachusetts for an air
quality permit to construct a cogenera-
tion power plant (the MATEP  power
plant) to be built near downtown Boston
to serve a hospital complex. As designed,
the new  plant would emit fewer pol-
lutants than the plant it  replaced and
would meet all existing ambient stan-
dards. Prior to 1977,  several  citizens'
groups had been battling Harvard's
expansion of the hospital  facilities into
their residential neighborhood.  The
proposed new construction served as
another rallying point for their opposi-
tion, and their goal became to stop the
plant rather than to search for alternative
sites or design configurations. TheirJ
tactics did focus,  however, on the'
scientific evidence and environmental
standards as well as legal maneuvers to
delay construction  and influence the
permit decision.
  The state's  modelers,  meanwhile,
focused on the federal standards. This
turned out to be too narrow a perspective
both politically and technically. Politi-
cally, the citizens' groups raised the
problem of short-term nitrogen dioxide
levels for which there is no federal
standard.  Technically, without  readily
available models to analyze the possibility
of an ozone limitation to the formation of
nitrogen  dioxide and the problem of
variability in diesel emissions at start-
up the state analysts tended to ignore
these problems. Such issues were
brought  up in the public hearings,
causing a delay while the problem was
analyzed, and  left the decision maker
faced with many uncertainties subject
to heated debate but no resolution.

  The Pittston  company has applied for
a permit to build an oil refinery in a rural
community in northern Maine near the
Canadian border and a national park.
The company redesigned the project
several times to reduce emissions so as 1
to meet the stringent PSD standards. "

 They were opposed by citizens of the
 surrounding community who felt  their
 livelihood (fishing) was threatened, as
 well as some influential U.S. Congress-
 men who had passed the PSD amend-
 ment to ensure that national parks
 retain their clean air. The EPA regional
 office reviewed the company's analyses
 and conducted their own in an effort to
 obtain a suitable project design. How-
 ever, they did not  look for alternative
 sites for such a project because that
 issue had been settled before on eco-
 nomic grounds and EPA was committed
 to supporting that decision.
  With the pressure to reach an agree-
 ment on the project discussions  with
 the regional administrator, the decision
 maker, was limited  to the question of
 when, not whether, the permit should
 be issued. Little discussion of the accu-
 racy of the models or what environmen-
 tal impacts may not have been measured
 took place. This was in part also due to
 technical issues. Although the  PSD
 standards are  easier to violate and
 demand more precise measurement,
 they appear to be of the same type as the
 other EPA standards. Because the deci-
 sion maker  was not familiar enough
 with the air quality models to make this
 distinction, he had to rely on the analysts.
 But the analysts did not raise this issue
 because they did not have other methods
 to draw on. In the end the citizens'
 groups seemed to have only a  limited
 impact on the decision making process.
 They were geographically separated
 from the EPA regional office in Boston
 and did not have access to modeling
 resources of their own. Many of the
 technical issues they did raise were not
 resolved within the time frame EPA had
 set for the decision.

  EPA's first proposal for an emissions
 standard for the copper smelter located
 in Anaconda, Montana was dismissed
 in court asan arbitrary figure which EPA
 had intended not as defendable but as
 bait for  further  discussion. Also, non-
ferrous  copper smelters were singled
out in the Clean Air Act as sources
particularly difficult to control and,
 hence, potentially able to be granted
exemptions from complying with stan-
dards. These two factors provided EPA
with the incentive it needed to commit
vast resources to modeling this copper
smelter. They  used a helicopter to
collect data and developed a model to be
used in complex mountainous terrain.
  It took over seven years to reach
agreement on an emissions limitation.
In general the state supported the
company, the largest employer in its
state. It was not until a new governor
was elected  and  the  company was
bought by the Atlantic Richfield Company
that a compliance schedule was agreed
upon. Thus, the agreement was due as
much to the change of governor and
company management as it was to the
extensive monitoring and validation
efforts of  EPA.  Nine months after the
agreement was signed and following a
lengthy, unsettled labor strike, however,
the company closed the plant citing
environmental,  health and safety rea-
sons as the cause.

  The Westvaco pulp and paper mill is
located  in rural  Maryland. In this case
the company collected monitoring data
and submitted modeling analyses to the
EPA regional office. The state again
supported a large employer within their
state. EPA did not have the resources to
do modeling of its own so it criticized the
company's analysis, pointing out that
the company's monitors were not sited
where standard violations were expected
to occur and suggesting several changes
in the technical analyses. The company
responded by submitting more data
from its misplaced monitors, building a
taller stack which would tend to disperse
rather than reduce the total amount of
pollution, and appealing to influential
U.S. Congressmen for support. There
was very little communication between
the parties except on the narrowtechni-
cal issues. Also, on several  issues,
including the  siting of monitors on
private  property and whether a taller
stack could or could not be used as a
dispersion technique, the EPA Head-
quarters office responsible for  setting
national  policy was silent. Thus, Westvaco
had no  incentive to upgrade either its
data or its analysis and a deadlock

Massachusetts' Sulfur
  Until the 1973 oil crisis Massachusetts'
regulation restricting the sulfur content
in fuel that could be burned by electric
utilities  was determined by the  availa-
bility of low sulfur fuel. But, as the price
began to  rise,  the state legislature
demanded a technical justification for
these regulations based on the impacts
on ambient air quality. Several interested
parties (the utilities, the state and EPA)
initiated modeling  analyses. These
analyses differed according tohoweach
party formulated the question to be
answered by the model based on their
own perspective. There was very little
effort made to coordinate these analyses.
In the end, the state based their decision
on their model, in part because no viola-
tions of the federal standards were
being monitored and in part because
EPA agreed that the  analysis was ac-
ceptable given the  limited time and

Ohio's State Implementation
  Ohio was the last large industrial
state to propose a State Implementation
Plan and could not agree on a control
plan for two large power plants located
near Lake Erie and owned by the Cleve-
land Electric Illuminating Company (CEI).
Thus, EPA was left to negotiate with CEI.
They both did modeling analyses based
on data collected by CEI from a monitor-
ing network which  EPA criticized as
inadequate. Meanwhile, the company
also sought help from Congress and the
White House. This aid came in the form
of a memorandum from the Council on
Wage and Price Stability which effec-
tively eliminated several control options
(installation of scrubbers and use of low
sulfur coal) leaving only the  noncom-
pliance option. The question of tall
stacks, long-range transport and acid
rain was not addressed because the
discussion of issues, as directed by EPA,
was never expanded beyond the narrow
local  impacts that the available air
quality models addressed while the
spectre of economic  hardship (without
technical analysis of the issue) won the

Connecticut's Transportation
Control Plan
  The state of Connecticut felt that their
oxidant problem was due to transport of
emissions downwind from sources
located in New York and  New Jersey.
The national policy of EPA was, however,
that each state was responsible for the
pollution that was monitored within its
own borders and their models did not
take the transport factor into account.
The state environmental agency at-
tempted to develop its own model but
ran into opposition from several quarters
including EPA. The state department of
transportation opposed the model be-
cause it did not like  the results which

indicated that many highway projects
would be rejected. The legislature and
the governor would not support any plan
requiring controls until EPA  acted to
make the other state implement controls.
When concessions were made to try to
draw up an acceptable plan, the envi-
ronmentalists opposed the changes. To
remove one obstacle  the state aban-
doned their model for an EPA approved
model. However, EPA had been threat-
ening sanctions if the state did not come
up with an acceptable transportation
control plan but never  carried out their
threats. Thus, the legislature felt free to
order another study and ensure that
nothing was done "prematurely,"
before other states in the region.

The San Francisco Air Quality
Maintenance Plan
  In  1975 the Association of Bay Area
Governments (ABAC)  was awarded  a
federal grant to develop an integrated
air quality, water quality and solid waste
management plan. It was the first grant
of its kind and set the stage for a sophis-
ticated and elaborate modeling analysis.
ABAC was also fortunate in having  a
private institution (the  Lawrence Liver-
more Laboratory) who raised funds from
outside sources to develop an air quality
model specific to the regions and useable
by the agency. The process followed by
ABAG in developing the integrated En-
vironmental Management Plan was the
one they had traditionally used for other
of the region's governmental programs.
It included the formation of task forces
at both the managerial and technical
staff level with a broad representation of
interests. The task forces were formed
at the  beginning of the process and
given the  responsibility of negotiating
an overall solution or agreement on the
facts. The primary aim of such a process
was public acceptability of the technical
data. It effectively removed the ability of
any one agency to exercise a veto over
the plan.
  The goals were limited. They included
improvement in air quality by the greatest
possible amount and compliance with
standards  at the earliest possible date
but subject to the condition that the plan
be  implementable.  This meant that
some measures, while capable of achiev-
ing immediate compliance with certain
standards, were rejected as politically
unacceptable or were relegated to
further discussion. But the relative suc-
cess of this process meant a concensus
was  reached among the technical ex-
perts that the best possible job had been
done. This in turn established technical
credibility in the eyes of the public, an
important step in establishing political
legitimacy for the plan.

Ozone Standard Revision
  When the question of whether or not
the national ambient air quality standard
for ozone ought to be revised arose, the
EPA Office of Research and Development
set up several task forces of scientists to
review both the health data and the air
quality models. The controversy sur-
rounding the standard revision decision
focused on the review of the health data
since these data would be the basis for
the decision. The review of the models'
capabilities was relatively non-contro-
versial because the models would only
be used later in implementing  the
standard and because the results of the
review suggested that the most flexible
of approaches to the use of models
should replace the existing regulation
which required the use of a linear
rollback type model.
  Although the Clean Air Act requires
an ambient standard to be based on
health  effects  alone,  a  Presidential
Executive Order requires the compilation
of the estimated costs of regulatory
actions. EPA's analysis of cost impacts
of the proposed standard included the
use of air quality models. At first a newly
developed model was used but as atten-
tion was focused on the results, the high
cost  of the standard, the linear rollback
model was substituted. It tended to give
lower cost estimates. The justification
for the change was that city specific
data  needed for the original model were
not available and that nationwide re-
duction estimates did not require a high
degree of accuracy. Thus, while the
choice of model may have been correct
for the problem as defined by EPA, the
choice was simplified by political con-

New Source Performance
  In  revising the new source perform-
ance standards for coal-fired power
plants EPA made a concerted effort to
develop a set of technical facts that the
various sides could agree on and which
could be used to defend the  standard.
That they were successful for the most
part  can be attributed to the early com-
mitment by the top management of both
EPA  and the U.S. Department of Energy
of not only time and resources but also a
commitment to a cooperative task force
arrangement for the actual development
of the technical information. They were
able to use a model of sufficient com-
plexity and flexibility for the situation
that was available and did not have to be
developed. EPA solicited public partici-
pation by both the utilities and environ-
mentalists from  the beginning of the
process and demonstrated their willing-
ness to change their analyses in response
to the public criticisms. Thus, the con-
scious management of  the process
helped EPA achieve its main aim, to
resolve as many technical issues as
possible  so as to focus  the political
debate on the political issues—whether
costs or emissions are more important—
rather than the technical issues—what
will the costs or emissions be.

  The problems encountered in these
case studies were  generally of four
  —technical and political constraints,
  —unresolved policy issues,
  —management by EPA of the decision
    process, and
  —conflicting institutional and organi-
    zational interests.
  The major technical problems included
a lack of verification studies for complex
topographic situations,  lack of input
monitoring  data, lack of sensitivity
analysis and output formats that did not
take into  account the broader question
of how air pollution control fits into the
energy and employment problems. The
political constraints included the energy
and employment  impacts of EPA's deci-
sions but also the interests and inter-
vention of outside parties such  as
industry and environmentalists. The
intervention by other parties became an
issue in those cases where the lobbying
activities  came toward the end of the
modeling process and delayed the
process while the modeler defended the
technical analysis. In those instances
when outside parties were encouraged
to participate from the beginning of the
process the  technical analysis was
changed  and improved in  response to
the lobbyists concerns.
  The policy problem centered on a lack
of directives from the federal EPA  on
interregional policy issues. Thus, we
saw that often the primary question
being debated was  what assumptions
as to wind speed or  worst case meteo-
rology to use for the model already
chosen. The question  of where to site
monitors to provide input data or which
model to use or whether any model can
adequately address the transport prob-

      was secondary and often never
 addressed. The answers to these ques-
 tions are technically harder but also,
 because of their national implications,
 often beyond the jurisdiction of the state
 or local agency overseeing the modeling.
 The  focus remained on the technical
 problems of the particular model with
 no mechanism to force a policy decision
 by the federal EPA.

   Several aspects of the management
 of the decision process led to an isola-
 tion of the technical analysis in the early
 stages of the process and a  lack of
 legitimacy or credibility for the analysis.
 No systematic effort at choosing the
 most appropriate model was made or, at
 a minimum, was  made clear to the
 public. Thus, the choice was not estab-
 lished as the best, given the state-of-
 the-art and resource constraints. The
 modeling analysis was then  done or
 critiqued by EPA but the early isolation
 left the modeler in a defensive posture,
 providing justification for the resources
 already spent, but lacking the resources
 or the desire to make changes. This lack
 of discussion of alternatives in turn  led
 to a  lack of credibility for the analysis.
|Those cases where public participation
"occurred before the model results were
 known were much more successful in
 getting technical issues resolved and
 the debate focused on the political

   In  each case study different  institu-
 tions and organizations had separate
 and conflicting interests in the impact of
 the decisions.  The clearest example is
 the opposing interests of industry and
 environmental interest groups. How-
 ever, within EPA the headquarters
 offices often have a different perspec-
 tive than the regional offices. Also, EPA
 and the states  differ on problems, such
 as transport, which impact differently
 on the local and national  levels. The
 consequence of these conflicting inter-
 ests  may be delay while they are heard
 although, again, a successful manage-
 ment of the  process will  use these
 interests to produce additional informa-
 tion and improved analyses.

   Not all of these factors will be under
 the control of the decision maker or the
 modeler. The recommendations of this
 report focus on those that can be con-
 trolled or changed (mainly the technical
 and managerial ones). They recognize
 and try to work within the others  as
 limitations or constraints on the decision
  The recommendations focus on what
steps can be taken to make the available
technical information, given the state-
of-the-art of air quality modeling and
the current legislative framework, more
useful to the decision maker. Thus, they
focus on technical planning and man-
agement changes and require new co-
operative efforts which recognize or-
ganizational and institutional conflicts.
1. Plan and require monitoring.
  Many of the ultimate decisions were
to require a better monitoring network,
but these decisions were made after a
delay of several years and controversial
court cases. Instead an appropriate
monitoring system should be required
as the first step in the decision process.
2. Detail guidelines for sensitivity anal-
ysis and output format when developing
a model.
  The inclusion of these activities in
model development rather than model
application will make sensitivity analyses
easier and the  model output more
flexible but,  more importantly, it will
force the modeler to develop a model
which is more likely to be relevant to the
policy (as opposed to research) setting.
3. Focus verification studies on actual
  EPA's  programs  to verify  models
should be based on problems discovered
through past applications of the models
and what problems the users expect
they will be called upon to address.
4. Include all interested parties from
the beginning.
  This could be done through early fact-
finding public hearings or the formation
of task forces. It appears to require extra
resources and time but the case studies
show that decisions take years and,
more importantly, that it is ccstly and
time consuming to have an issue raised
late in the process when the  issue could
and should have been dealt with earlier.
5. Establish criteria and boundaries for
the technical analysis for each decision.
  If such questions as what are the
broader  issues,  what is the range of
policy issues, and if modeling should be
done  at all are asked before choosing
which model, which data or which
assumptions, it is more likely that the
decision  will  not founder on technical
questions but instead can be focused on
the policy issues.

6. Establish agreement on state-of-the-
art and assumptions before the techni-
cal results are presented.
  Early  in the process the discussion
should focus on what is the appropriate
model given the available resources. In
this way it is harder to choose a model
because of what results it gives rather
than because it is the right model for the
situation. How to negotiate such agree-
ments will come through experience as
there are few examples of such attempts,
much less successful outcomes.
7. Establish joint industry/EPA model-
ing studies.
  This could help force agreement early
on the technical assumptions. Other
interests should also be included but the
major resources would be from industry
and EPA.
8. Explain decision and alternatives
considered in detail.
  To establish legitimacy for the process
the basis for the decision has to be made
public. This also encourages the broader
consideration of alternatives during the
decision process.

Catherine G. Miller is with the John F. Kennedy School of Government, Harvard
  University. Cambridge, MA 02138.
Kenneth Demerjian is the EPA Project Officer (see below).
The complete report, entitled "Case Studies in the Application of Air Quality
  Modeling in Environmental Decision Making," (Order No. PB 81 -213 233;
  Cost: $9.50, subject to change) will be available only from:
        National Technical Information Service
        5285 Port Royal Road
        Springfield, VA 22161
        Telephone:  703-487-4650
The EPA Project Officer can be contacted at:
        Environmental Sciences Research Laboratory
        U.S. Environmental Protection Agency
        Research Triangle Park, NC 27711
                                                                                   * US GOVERNMENT PRINTING OFFICE. 1981 -757-012/7280

United States
Environmental Protection
Center for Environmental Research
Cincinnati OH 45268
Postage and
Fees Paid
EPA 335
Official Business
Penalty for Private Use $300
      PS   0000329
      U  S  EtftflK  PROTtCTIUM
      KEG ION  5 I, I BRAKY
      230  S DEARBURN  STRfcEI
      CHICAGO II. 60604